2. Theoretical expectations

Economic voting research (Lewis-Beck and Stegmaier, Reference Lewis-Beck and Stegmaier2019) finds that negative economic events harm those in power, as voters use these events to update their economic perceptions and reassess government performance. A massive and unexpected economic shock, such as the Lehman crash, is likely to be a strong political signal to European voters for at least two reasons. First, even before policy implementation, economic news shocks are politically consequential (Ciobanu, Reference Ciobanu2024). Second, events taking place in the US impact European public opinion (Minkus et al., Reference Minkus, Deutschmann and Delhey2019; Delis et al., Reference Delis, Matakos and Xefteris2020). However, there are also reasons to expect that a foreign economic shock will not follow the economic voting logic, at least in the short term. The shock’s external nature is likely to break the connection between economic events and political evaluations, as voters cannot reasonably attribute responsibility (Iyengar, Reference Iyengar1991) for these far-off, albeit severe, events to their national political leaders. Our key claim is that, as voters become more pessimistic about the national economy immediately following an external economic shock, they will trust their national leaders more, not less.

Our argument, following Minkus et al. (Reference Minkus, Deutschmann and Delhey2019) findings, is that the Lehman collapse could be perceived by the public as an economic threat in Europe due to the uncertainty it produces and its potential to destabilize the global economy. Therefore, the threat is likely an economic one: the event is thus hypothesized to worsen the Europeans’ perceptions of the national economic situation. Confronted with economic news about the probability of a global economic meltdown and given the importance of the US economy and its strong ties with Europe, and the potential spread of the crisis to their country, voters are likely to worry about the state of their national economy.

The foreign economic shock is likely to bring voters closer to their national leaders in the short term. External threats—military, terrorist, or public health—tend to increase support for incumbents (Hetherington and Nelson, Reference Hetherington and Nelson2003; De Vries et al., Reference De Vries, Bakker, Hobolt and Arceneaux2021). Scholarship also finds that international economic developments affect incumbent support (Campello and Zucco, Reference Campello and Zucco2016). We expect a similar pattern following the Lehman bankruptcy: as a response to the external economic threat, the voters are likely to rally around the political establishment (Minkus et al., Reference Minkus, Deutschmann and Delhey2019), which, after the shock, will increase the satisfaction with the national government. As already shown (Delis et al., Reference Delis, Matakos and Xefteris2020; De Vries et al., Reference De Vries, Bakker, Hobolt and Arceneaux2021), in times of (perceived) crisis, voters will opt for greater political stability and support those who have the tools to reduce uncertainty.

Our argument aligns with Hellwig and Samuels (Reference Hellwig and Samuels2007), who find that global exposure weakens economic voting, and with Duch and Stevenson (Reference Duch and Stevenson2010), who show that open economies experience weaker economic voting due to greater vulnerability to shocks. We build on Albertson and Gadarian (Reference Albertson and Gadarian2015), who argue that uncertainty triggers anxiety, prompting citizens to seek information and trust the government (Albertson and Gadarian, Reference Albertson and Gadarian2015, 10). Since the Lehman shock was external, citizens do not assign causal responsibility to the government (Iyengar, Reference Iyengar1991) and may instead increase trust when it “can offer solutions to threats” (Albertson and Gadarian, Reference Albertson and Gadarian2015, 16). Following the “rally-round-the-flag” logic (Hetherington and Nelson, Reference Hetherington and Nelson2003), rising trust may result from muted opposition criticism and a positive response from opposition supporters—a mechanism we explore.

3. Case: The Lehman Brothers collapse

The Lehman Brothers bankruptcy was a key moment in bringing about the worst financial and economic crisis since the 1920s (see Sorkin (Reference Stewart2009), Stewart (Reference Sorkin2009), and Swedberg (Reference Swedberg2010) for detailed accounts of the events). As Blinder (Reference Blinder2009) argues, “everything fell apart after Lehman. [...] After Lehman went over the cliff, no financial institution seemed safe. So, [...] the economy sank like a stone.” Economists agree that the bankruptcy transformed the “subprime crisis” that broke out in August 2007 and was mainly a financial crisis restricted to the US and UK into a “global crisis” (Baldwin, Reference Baldwin2009, 9). The Lehman outcome “shocked the global financial community”, since they had assumed that “no major financial institution would be allowed to go under” (Baldwin, Reference Baldwin2009, 9). As per Paulson (Reference Paulson2008), the US Secretary of the Treasury, the consequences of the event were massive: “the financial system essentially seized up and we had a system-wide crisis. Credit markets froze and banks substantially reduced interbank lending. Confidence was seriously compromised throughout our financial system.” Although the bankruptcy became likely over the weekend before 15 September and the US financial system started to prepare for the event (Stewart, Reference Stewart2009), those in charge of the US markets “did not think that the fall of the Lehman would set off the kind of panic it did”: according to Swedberg (Reference Swedberg2010), the Lehman’s bankruptcy “set off a panic that would end up by threatening not only the US financial system but also the global financial system.”

The fallout of Lehman spread internationally immediately and the financial effects were massive (Swedberg, Reference Swedberg2010). In the US, the Dow Jones declined by 4.4% on the announcement day, its “largest drop” (Swedberg, Reference Swedberg2010, 93) following the 9/11 terrorist attacks. The same happened in Europe and Asia, where the stock markets fell by 2-4%, and they continued to plunge in the following days (Stewart, Reference Stewart2009). Like the US, the EU and the UK rushed to implement large public guarantees and interventions to stabilize and calm the markets. Following the Lehman collapse, “world trade experienced a sudden, severe and synchronized collapse [...]—the sharpest in recorded history and deepest since WWII” (Baldwin, Reference Baldwin2009, 1). The chronology of the events show how fast the effects were felt worldwide and how much pressure there was on the governments to react immediately.Footnote ¹ The Lehman bankruptcy announcement was truly a massive, unexpected, and consequential economic shock with global ramifications.

4. Research design

We leverage the fact that the fourth wave of the ESS was in the field at the time of the Lehman Brothers crash (15 September 2008). This enables us to apply the Unexpected Events During Survey Design (UESD) framework (Muñoz et al., Reference Muñoz, Falcó-Gimeno and Hernández2020), where a sudden, salient event randomly splits respondents into treatment and control groups based on interview timing. Our identification strategy is similar to the one used by Minkus et al. (Reference Minkus, Deutschmann and Delhey2019). We also build on Podlich et al. (Reference Podlich, Schnabel and Tischer2017), who argue that the Lehman default can be considered an unexpected and exogenous shock to the German banking system. We apply this logic to the six countries (Switzerland, Germany, Spain, United Kingdom, Netherlands, and Norway) for which we have pre- and post-bankruptcy observations in our ESS sample ( $N = 12,825$) and that offer good geographical variation.

Our key dependent variable is satisfaction with the national government, measured on a 0–10 scale. We also test whether the event influences the satisfaction with the national economy (measured on the same scale), as the economic nature of the shock would suggest. Our main explanatory variable registers whether someone was interviewed before 15 September or after. In our sample, 12% of the respondents were interviewed before the announcement. The treatment is a bundled one, as it consists of all the events following the collapse; thus, the results should be interpreted as the effects of the crisis triggered by the shock. We estimate OLS regressions, in which we regress the outcomes on the key independent variable; we show models with and without controls. As controls, we employ the following variables: sex, age, education, urban residence, marital status, subjective income, being an ethnic minority, country citizenship, employment status, turnout in the recent election, and incumbent support in the recent election. Country fixed effects are also included. We discuss models for the entire sample, but we also show how the effect evolves when we focus on those interviewed a number of days (from 1 to 60) before and after the announcement. Descriptive statistics for all variables are presented in Supplementary material, Appendix A.

We discuss the credibility of the research design. First, to provide further evidence for the unexpectedness of the event, we employ Google Trends using “Lehman Brothers” as the search term. The results are visualized in Figure B.1 (Appendix B) and show, for all countries, that the public’s interest in the term is at the minimum before the event but explodes and reaches its highest point in the post-announcement week. Similar trends are observed with the “crisis” search term (Figure B.2). Second, to increase confidence in our design, we show equivalence-based balance tests (Hartman and Hidalgo, Reference Hartman and Hidalgo2018) for the control variables mentioned above. We find very good balance for our sample (Figure B.3, Appendix B). These two tests validate the unexpected event logic.

5. Findings

We first analyze whether the bankruptcy announcement led to less satisfaction with the national economy. This represents a precondition for our argument: for political effects to emerge, an increase in economic threat should be visible. In the post-announcement period, the satisfaction with the national economy deteriorated by half a point (0.53) on a 0–10 scale (Table 1, models 1–2). Moreover, the economic dissatisfaction increases over time; as voters are more exposed to economic news about the event and its likely consequences, their reaction becomes stronger.Footnote ² The deterioration of the economic sentiment at the national level is present for each country, although the estimates are sometimes imprecise (Appendix C).

Table 1. Satisfaction with the national economy and government (all countries)

Note: $^{***}p \lt 0.01$; $^{**}p \lt 0.05$; $^{*}p \lt 0.1$. OLS estimations with standard errors in parentheses. Dependent variables for the models: satisfaction with the national economy (0–10)—models 1–2; satisfaction with the national government (0–10)—models 3–4. All models use survey weights.

We zoom in on the effect evolution. Here, we illustrate how the coefficient changes as we incrementally increase the number of days before and after the event from 1 to 60. The ESS data collection starts on 21 August 2008, so we have 25 days before the shock. We add control variables and country dummies. As observed in Figure 1, the satisfaction with the national economy generally decreases over time, from an effect size of $-$0.27 in the window of four days around the event to $-0.41$ 60 days post-shock.Footnote ³ The results are very similar without controls (Figure C.1). We also employ a different estimation strategy, in which we keep all the pre-event observations and we increment the number of days post-event; the results (with and without controls) are substantively very similar to Figure 1—Figures C.2–C.3 in Appendix C. We demonstrate that this dynamic is present in all countries, although the coefficients are not always precisely estimated (Figures C.4–C.5). We also conduct a placebo test: we focus on the observations collected in the pre-treatment period and we consider a hypothetical treatment day (1 September 2008); reassuringly, a clear deterioration of the economic sentiment is absent (Appendix H). Moreover, the shock does not affect satisfaction with the state of education and health services (Appendix G), two placebo outcomes. The shock does not impact the likelihood of becoming unemployed or of not having enough money for the household’s necessities (Appendix D). The crisis is internalized sociotropically, not egotropically.

Note: Each coefficient comes from a separate estimation using equal numbers of days before and after the event (e.g., “5” on the x-axis reflects data from 5 days before and after the announcement). Data collection begins on 21 August 2008.

Figure 1. Satisfaction with the national economy (all countries, coefficient plot).

Next, we investigate the consequences of the Lehman debacle on the satisfaction with the national government. In the aftermath of the crash, in Europe, the government satisfaction increased by 0.14–0.25 on a 0–10 scale (Table 1, models 3–4).Footnote ⁴ Again, the effect sizes tend to increase as time passes (Table E.1). Additional tests reveal that the effect is stronger among those who did not vote for the incumbent (Table I.1). Moreover, sentiment analysis of parliamentary speeches in four countries suggests opposition parties did not adopt a more negative tone after the shock (Figure I.4). Both findings support the opinion leadership mechanism from the “rally-round-the-flag” literature (Hetherington and Nelson, Reference Hetherington and Nelson2003). The rally effect appears in four countries—statistically significant in the UK and Netherlands, but not in Spain and Norway (Appendix E). In line with rally mechanisms, past opposition voters respond most strongly in the UK and Netherlands (Table I.1). These are also the most globally integrated countries in our sample, based on financial, trade, and informational globalization scores (Appendix J) from the KOF Index (Gygli et al., Reference Gygli, Haelg, Potrafke and Sturm2019), which may explain their heightened sensitivity to US events. In contrast, Germany and Switzerland had either a grand coalition or broad power-sharing at the time, likely dampening opposition responses. We tentatively conclude that our argument applies best in globally integrated countries with clear government–opposition divides.

We follow the same approach as in Figure 1 to assess the effect’s dynamics. In Figure 2, we see that the effect is statistically significant starting 14 days after the shock (0.16), it increases to 0.2 25 days after the event, and is 0.29 60 days post-event. This gradual emergence aligns with evidence indicating that the effects of economic shocks on political attitudes and behavior unfold over time, rather than manifesting immediately, and are shaped by increasing media attention to the shock (Ciobanu, Reference Ciobanu2024). In line with this, earlier research shows that media coverage influences economic perceptions and that such updating processes occur gradually (Doms and Morin, Reference Doms and Morin2004). Comparable delayed dynamics have been documented in response to terrorist attacks (Epifanio et al., Reference Epifanio, Giani and Ivandic2023) and court rulings (Bridgman et al., Reference Bridgman, Ciobanu, Erlich, Bohonos and Ross2021). Four pieces of evidence support this trend: (1) a gradual rise in Google Trends searches for “crisis” (Figure B.2); (2) an increasing share of media articles mentioning Lehman, the economy, or crisis across six countries, based on Dow Jones Factiva (Figure I.1); (3) a similar upward trend in parliamentary mentions of these terms in the UK, Netherlands, Germany, and Spain, using the ParlSpeech dataset (Rauh and Schwalbach, Reference Rauh and Schwalbach2020) (Figure I.2); and 4) a parallel pattern in prime ministerial speeches in the UK, Netherlands, and Spain (Schumacher et al., Reference Schumacher, Berk, Pipal, Kantorowicz, Schoonvelde, Traber and de Vries2020) (Figure I.3). Substantively similar results to those in Figure 2 are obtained without controls (Figure E.1) and when using a different estimation strategy that considers all pre-event observations for all estimations (Figures E.2–E.3). Country-level estimations (Figures E.4–E.5) confirm that the changes are particularly present in the United Kingdom and the Netherlands, with the expected positive impact also identified in Spain and Norway. A placebo test with a hypothetical bankruptcy date (1 September) and restricted to pre-treatment observations finds a non-significant effect (Appendix H). No impact is detected for several placebo outcomes (satisfaction with democracy, partisanship, reported turnout and vote in the last election)—Appendix G.

Note: Each coefficient comes from a separate estimation using equal numbers of days before and after the event (e.g., “5” on the x-axis reflects data from 5 days before and after the announcement). Data collection begins on 21 August 2008.

Figure 2. Satisfaction with the national government (all countries, coefficient plot).

We replicate our main results for all 27 European Union members with a different dataset collected shortly before and after the shock (Appendix L). While perceptions of the national economy deteriorate by about 5–6 points, trust in the national government increases by 2 points. Moreover, applying the UESD framework to ANES data, we find, as expected, that the Lehman shock affected perceptions of the national economy in the US but did not favor the incumbents (Appendix M).

6. Conclusion

Despite the economic importance of the Lehman bankruptcy, little is known about its short-term political consequences. By treating the crash as an unexpected and exogenous event in Europe, we investigate how it affected political attitudes. We argue that, faced with an external economic threat, voters will rally behind the government despite worsening economic perceptions. We find that while the Lehman shock increased dissatisfaction with the national economy in Europe, it also led to a gradual increase in satisfaction with the national government post-event. We document that rising media and political attention coincided with a rally effect fueled by past opposition voters and muted opposition elites. With a different dataset, we show that these findings apply to all 27 European Union members. Moreover, when applying the UESD logic to the US case, we observe that, while the European and American voters shared similar evaluations of the national economy, they had different political assessments. This is to be expected given the origin of the shock and consistent with our argument. Our work contributes to research on the political impact of economic shocks (Margalit, Reference Margalit2019; Baccini et al., Reference Baccini, Ciobanu and Pelc2025), including the post-2008 crisis (Hernandez and Kriesi, Reference Hernandez and Kriesi2016), and on the political effects of foreign events (Campello and Zucco, Reference Campello and Zucco2016; Minkus et al., Reference Minkus, Deutschmann and Delhey2019; De Vries et al., Reference De Vries, Bakker, Hobolt and Arceneaux2021). Moreover, we show that external economic shocks modify the economic voting logic by making economically worried voters more satisfied with their national governments, not less. It is not only global economic exposure (Hellwig and Samuels, Reference Hellwig and Samuels2007; Duch and Stevenson, Reference Duch and Stevenson2010), but also external shocks that constrain economic voting.

The findings reveal two broader implications: first, they highlight the limits of democratic accountability, as governments benefit from events beyond their control; and second, the boost in government support could in fact be good news, as it could help the decision-makers better navigate a potential crisis in the absence of a hostile public. This is highly relevant, as it increases governments’ ability to respond to immediate threats from economic and financial crises, as well as the economic impacts of pandemics and war, as seen in recent years.